Elon Musk, the CEO of Twitter, threatened to sue Microsoft on Wednesday, accusing the company of illegally using Twitter's data in order to train the artificial intelligence model it uses to help users interact with it.
It was reported that Musk would issue a tweet about dropping Twitter from Microsoft's social media advertising platform, which makes it easy for ad buyers to manage their social media accounts in one place after the news was reported by Mashable and other publications.
There have been illegal training sessions using Twitter data, Musk tweeted during the training period. "Lawsuit time!".
Oftentimes, Musk, who also serves as the CEO of SpaceX and Tesla, tweets about plans that never materialize, and up until now there has never been a lawsuit filed over it. Twitter's press line didn't respond meaningfully to an inquiry for comment, and Microsoft declined to comment on its side as well.
It is becoming increasingly apparent in the era of generative AI that data ownership has become a fraught battleground. Musk's threat is just one example. In a fast-paced technological world where big tech companies are trying to develop cutting-edge artificial intelligence models like OpenAI's GPT, data owners are trying to halt these efforts or to charge for their use.
In an unusually structured deal last year, Microsoft invested $10 billion in OpenAI, an unusually structured venture that involved developing its own large language models and selling access to OpenAI's models. Previously, Musk served as a co-founder of OpenAI, and he has complained recently of the company's transition from being a nonprofit to becoming a highly valuable business influenced by Microsoft after it had moved from a nonprofit model to a highly valuable business.
Training data from social networks is particularly valuable to LLMs such as GPT because it allows us to capture informal, back-and-forth conversations between users. The data from social networks is far more relevant to the training process than that gathered from websites like Reddit, StackOverflow, and Twitter.
Increasingly, it is appearing that the owners of the data are making demands as these new AI models are being adopted beyond the research labs and universities.
Reddit, for example, announced earlier this week that it would charge companies for access to its programming interface, which is used to feed the conversations among Redditors into artificial intelligence software that is designed to teach the language of humans. In response to the viral video of a song in which allegedly used artificial intelligence to imitate the rapper Drake, Universal Music Group stated this week that such a training of artists' music would be a breach of the company's agreements and a violation of copyright law.
It has also been reported that Getty Images, a stock photo database, has sued Stable Diffusion for using its content to train its AI image generator, alleging that the company copied Getty Images' content.
In December, Musk said Twitter would temporarily block OpenAI access to its database, and he also announced that he will be building his own large language model as part of one of his companies, TruthGPT, which will be funded by Musk.
As a leading independent research provider, TradeAlgo keeps you connected from anywhere.